基于神经网络的不确定性数据流异常检测系统设计

Design of anomaly detection system for uncertain data flow based on neural network

受非线性变化数据影响,导致数据流异常检测结果不精准,为此设计了基于神经网络的不确定性数据流异常检测系统。采集时间窗口数据,计算离散值,将计算结果存入综合数据库。提取不确定性数据流异常特征,结合神经网络检测数据流异常情况。构建原始数据流序列和不确定性数据流序列,并以此为基础构建检测模型。引入递推算法,结合Lasso回归分析方法剔除非线性变化数据,分析不确定性数据的异常特性,通过神经网络锁定异常数据流,获取检测结果。由实验结果可知,该系统可将数据拟合在理想值附近,且样本数据在实际值上下限范围内,能够获取精准的检测结果。

Due to the influence of nonlinear change data,the data flow anomaly detection results are inaccurate.Therefore,an uncertain data flow anomaly detection system based on neural network is designed.Collect time window data,calculate discrete values,and store the calculation results in the comprehensive database.The anomaly features of uncertain data flow are extracted,and the anomaly of data flow is detected with neural network.The original data flow sequence and the uncertain data flow sequence are constructed,and the detection model is built based on them.Recursive algorithm is introduced,combined with Lasso regression analysis method to eliminate nonlinear change data,analyze abnormal characteristics of uncertain data,lock abnormal data flow through neural network,and obtain detection results.The experimental results show that the system can fit the data near the ideal value,and the sample data is within the upper and lower limits of the actual value,which can obtain accurate detection results.